Gaurav Verma

Computer Science Ph.D. student at Georgia Tech | Interested in Multimodal Machine Learning and Natural Language Processing


[ curriculum vitae (pdf) ]


I am a second-year Ph.D. student in Computer Science at Georgia Tech, where I am advised by Prof. Srijan Kumar. I am interested in developing robust and trustworthy machine learning methods that can fuse information from different modalities to solve problems that impact the well-being of individuals and society. My specific interests include: learning from multimodal data, adversarial machine learning, natural language processing, and computational social science. [CV]

I completed my undergraduate studies at the Indian Institute of Technology Kanpur. At IIT Kanpur, I worked with Prof. Tanaya Guha on learning modality-independent representations for affective analysis and retrieval of multimedia. Before starting my doctoral studies at Georgia Tech, I was at Adobe Research, where I worked on a wide range of research projects, like stylistic analysis and generation, multimodal content synthesis [TechCrunch], query construction, and action recommendation. Please see the Publications page for more details on published papers and patents.

Apart from research, I love to spend most of my time reading books (📚). I also have some affinity for sports – basketball (🏀) and table tennis (🏓). The quickest way to get in touch would be through email (📧). I am also somewhat active on Twitter. More details are available in the Contact section.

Academic Service: Program Committee Member/Reviewer for:
ICLR 2022, TheWebConf (WWW) 2022, EMNLP 2021, ACL-IJCNLP 2021, ICWSM 2021 (💐 Best Reviewer Award), ICWSM 2022, ACII 2021, CODS-COMAD 2022, and ECIR 2020.

Recent Updates

Mar, 2022 | Our work on using multimodal learning to overcome the language disparity in online content classification has been accepted to ICWSM 2022 🎉 ! [pdf coming soon]
Jan, 2022 | Happy to share that our work on online ban evasion has been accepted to The Web Conference 2022 🎉 ! [pdf] [dataset]
Oct, 2021 | My collaborators from Adobe Research and I received the Best Paper Runner-up Award 💐 at WISE 2021 for our work on detecting document versions [pdf, image]
June, 2021 | I was one of the winners of ICWSM-2021 Best Reviewer Award 💐
Dec, 2020 | Our work on consuming linear media (like videos) in a non-linear fashion using multimodal fragments has been accepted at IUI 2021! [pdf] [video]
Sep, 2020 | EMNLP 2020: Check out our Findings paper on using reinforcement learning for generating stylized text [arXiv] and our system description for Task 2 at W-NUT [arXiv].
July, 2020 | TechCrunch talks about #ProjectSnippets in their blog! Read it here. This tool is based on our IUI'20 paper on Generating Need Adapted Multimodal Fragments [pdf].
June, 2020 | We participated in ICWSM'20 Safety Data Challenge. Here's our paper.
Mar, 2020 | Check out our work on generating multimodal fragments being presented at this year's Adobe Summit: YouTube video! Here's what the community thinks about #ProjectSnippets! [News]
Feb, 2020 | Our exploration on estimating the causal impact of stylistic attributes on a targeted goal has been accepted at WWW 2020 as a poster! Here's the paper.
Dec, 2019 | Our work on "Generating Need-Adapted Multimodal Fragments" has been accepted at IUI 2020! Check out the paper.
Dec, 2019 | Our work on "Using Image Captions and Multitask Learning for Recommending Query Reformulations" has been accepted at ECIR 2020. Here's the paper on arXiv.
Nov, 2019 | Our work on "Adapting Language Models for Non-Parallel Author-Stylized Rewriting" has been accepted as a full paper at AAAI 2020. Here's the paper on arXiv.
Aug, 2019 | Adobe asked me a few questions on completing one year at Adobe Research [AdobeLife]
Mar, 2019 | Work on command recommendation accepted at UMAP 2019 [project page]
Feb, 2019 | Work on multimodal affective correspondence accepted at ICASSP 2019 [page]